Abstract
When two masked objects are presented in close temporal succession correct identification of the first target (T1) leads to a severe impairment in identification of the second (T2). Although the precise mechanism that underlies this ‘attentional blink’ (AB) has been debated, there is consensus that the AB reflects the temporal distribution of attention and that it occurs because of a single resource or capacity limitation that is occupied with T1 processing. Recently, it has been proposed that there are multiple resource channels available for processing selected information. According to this view, the severity of the AB is dependent on the channels required by T1 and T2, such that the AB will be less severe if the target objects require different channels for processing. The purpose of the present experiment was to test whether this reduction in the AB depends on whether the dimension on which the targets are similar is task-relevant. Participants were presented faces in RSVP. Face gender, emotional valence (neutral and fearful), and T1–T2 lag (240ms and 560ms) was manipulated factorially. Task-relevance was manipulated by having participants report the gender or valence of T1 and T2. The AB was most severe when T1 and T2 were the same task relevant dimension relative to when they were not. This pattern did not hold for the task irrelevant dimension. The results support the conclusion that the relationship between T1 and T2 can modulate the AB, but only when that relationship is task-relevant.